Performance Based Matchmaking on Grid

Grid Technologies supply users with high computational and storage resources to execute demanding applications. To this end, Grid environments must provide query and discovery tools, able to select the most suitable resource(s) satisfying application requirements. A description of application and resources, grounded on a common and shared basis, is therefore crucial to favour an effective pairing. A viable criterion to match demand (job) with supply (computational resource) is to characterize resources by means of their performance evaluated through benchmarks relevant to the application. We introduce GREEN, a distributed matchmaker, based on a two-level benchmarking methodology. GREEN facilitates the submission of jobs to the Grid, through the specification of both syntactic and performance requirements, independently of the underlying middleware and thus fostering Grid interoperability.